Pull the plug on bad AI.
unplug is an LLM defense platform. Scan untrusted text for prompt injection,
destructive actions, and data leakage. Get evidence-based findings and span-level
redaction — not binary blocking.
Capabilities
- Prompt injection detection with normalization pipeline
- Destructive action blocking (SQL, shell, git, file operations)
- Data leakage prevention (API keys, PII, prompt leaks)
- Free Python SDK — pip install unplug-ai
- Hosted API at api.unplug-ai.org
Links
Enable JavaScript for the full interactive site.